Upgrade to Encode 1.63.
[p5sagit/p5-mst-13.2.git] / ext / Encode / Encode.xs
CommitLineData
85982a32 1/*
10c5ecbb 2 $Id: Encode.xs,v 1.41 2002/04/27 18:59:50 dankogai Exp $
85982a32 3 */
4
fcf2db38 5#define PERL_NO_GET_CONTEXT
2c674647 6#include "EXTERN.h"
7#include "perl.h"
8#include "XSUB.h"
2f2b4ff2 9#define U8 U8
10#include "encode.h"
10c5ecbb 11
12# define PERLIO_MODNAME "PerlIO::encoding"
0ab8f81e 13# define PERLIO_FILENAME "PerlIO/encoding.pm"
aae85ceb 14
85982a32 15/* set 1 or more to profile. t/encoding.t dumps core because of
16 Perl_warner and PerlIO don't work well */
0ab8f81e 17#define ENCODE_XS_PROFILE 0
39cf9a5e 18
85982a32 19/* set 0 to disable floating point to calculate buffer size for
20 encode_method(). 1 is recommended. 2 restores NI-S original */
0ab8f81e 21#define ENCODE_XS_USEFP 1
39cf9a5e 22
fcf2db38 23#define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
2f5768b8 24 Perl_croak(aTHX_ "panic_unimplemented"); \
4a83738a 25 return (y)0; /* fool picky compilers */ \
87714904 26 }
85982a32 27/**/
011b2d2f 28
7e9a885a 29UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
30UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
33af2bc7 31
b0b300a3 32void
aa0053b7 33Encode_XSEncoding(pTHX_ encode_t * enc)
2f2b4ff2 34{
aa0053b7 35 dSP;
36 HV *stash = gv_stashpv("Encode::XS", TRUE);
37 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
38 int i = 0;
39 PUSHMARK(sp);
40 XPUSHs(sv);
41 while (enc->name[i]) {
42 const char *name = enc->name[i++];
43 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
44 }
45 PUTBACK;
46 call_pv("Encode::define_encoding", G_DISCARD);
47 SvREFCNT_dec(sv);
2f2b4ff2 48}
49
aa0053b7 50void
51call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
52{
85982a32 53 /* Exists for breakpointing */
aa0053b7 54}
67e989fb 55
85982a32 56
2f2b4ff2 57static SV *
aa0053b7 58encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
85982a32 59 int check)
2f2b4ff2 60{
aa0053b7 61 STRLEN slen;
62 U8 *s = (U8 *) SvPV(src, slen);
3aececda 63 STRLEN tlen = slen;
64 STRLEN ddone = 0;
65 STRLEN sdone = 0;
39cf9a5e 66
3c49ab08 67 /* We allocate slen+1.
85982a32 68 PerlIO dumps core if this value is smaller than this. */
3c49ab08 69 SV *dst = sv_2mortal(newSV(slen+1));
85982a32 70 U8 *d = (U8 *)SvPVX(dst);
71 STRLEN dlen = SvLEN(dst)-1;
72 int code;
73
74 if (!slen){
75 SvCUR_set(dst, 0);
76 SvPOK_only(dst);
77 goto ENCODE_END;
78 }
79
80 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
81 {
82 SvCUR_set(dst, dlen+ddone);
83 SvPOK_only(dst);
0b3236bb 84
85982a32 85 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
86 break;
87 }
88 switch (code) {
89 case ENCODE_NOSPACE:
90 {
91 STRLEN more = 0; /* make sure you initialize! */
92 STRLEN sleft;
93 sdone += slen;
94 ddone += dlen;
95 sleft = tlen - sdone;
fcb875d4 96#if ENCODE_XS_PROFILE >= 2
85982a32 97 Perl_warn(aTHX_
98 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
99 more, sdone, sleft, SvLEN(dst));
fcb875d4 100#endif
85982a32 101 if (sdone != 0) { /* has src ever been processed ? */
39cf9a5e 102#if ENCODE_XS_USEFP == 2
85982a32 103 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
104 - SvLEN(dst);
39cf9a5e 105#elif ENCODE_XS_USEFP
85982a32 106 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
39cf9a5e 107#else
85982a32 108 /* safe until SvLEN(dst) == MAX_INT/16 */
109 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
39cf9a5e 110#endif
39cf9a5e 111 }
85982a32 112 more += UTF8_MAXLEN; /* insurance policy */
113 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
114 /* dst need to grow need MORE bytes! */
115 if (ddone >= SvLEN(dst)) {
116 Perl_croak(aTHX_ "Destination couldn't be grown.");
117 }
118 dlen = SvLEN(dst)-ddone-1;
119 d += ddone;
120 s += slen;
121 slen = tlen-sdone;
122 continue;
123 }
124 case ENCODE_NOREP:
125 /* encoding */
0ab8f81e 126 if (dir == enc->f_utf8) {
85982a32 127 STRLEN clen;
128 UV ch =
0ab8f81e 129 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
b0b300a3 130 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
85982a32 131 if (check & ENCODE_DIE_ON_ERR) {
132 Perl_croak(
0ab8f81e 133 aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d",
85982a32 134 ch, enc->name[0], __LINE__);
135 }else{
136 if (check & ENCODE_RETURN_ON_ERR){
137 if (check & ENCODE_WARN_ON_ERR){
138 Perl_warner(
139 aTHX_ packWARN(WARN_UTF8),
0ab8f81e 140 "\"\\N{U+%" UVxf "}\" does not map to %s",
85982a32 141 ch,enc->name[0]);
142 }
143 goto ENCODE_SET_SRC;
144 }else if (check & ENCODE_PERLQQ){
0ab8f81e 145 SV* perlqq =
e4da7dce 146 sv_2mortal(newSVpvf("\\x{%04"UVxf"}", ch));
b2704119 147 sdone += slen + clen;
85982a32 148 ddone += dlen + SvCUR(perlqq);
149 sv_catsv(dst, perlqq);
af1f55d9 150 }else if (check & ENCODE_HTMLCREF){
151 SV* htmlcref =
7e19fb92 152 sv_2mortal(newSVpvf("&#%" UVuf ";", ch));
af1f55d9 153 sdone += slen + clen;
154 ddone += dlen + SvCUR(htmlcref);
155 sv_catsv(dst, htmlcref);
156 }else if (check & ENCODE_XMLCREF){
157 SV* xmlcref =
158 sv_2mortal(newSVpvf("&#x%" UVxf ";", ch));
159 sdone += slen + clen;
160 ddone += dlen + SvCUR(xmlcref);
161 sv_catsv(dst, xmlcref);
0ab8f81e 162 } else {
85982a32 163 /* fallback char */
164 sdone += slen + clen;
0ab8f81e 165 ddone += dlen + enc->replen;
166 sv_catpvn(dst, (char*)enc->rep, enc->replen);
b2704119 167 }
0ab8f81e 168 }
b2704119 169 }
85982a32 170 /* decoding */
0ab8f81e 171 else {
85982a32 172 if (check & ENCODE_DIE_ON_ERR){
173 Perl_croak(
fc17bd48 174 aTHX_ "%s \"\\x%02" UVXf
175 "\" does not map to Unicode (%d)",
85982a32 176 enc->name[0], (U8) s[slen], code);
177 }else{
178 if (check & ENCODE_RETURN_ON_ERR){
179 if (check & ENCODE_WARN_ON_ERR){
180 Perl_warner(
181 aTHX_ packWARN(WARN_UTF8),
fc17bd48 182 "%s \"\\x%02" UVXf
183 "\" does not map to Unicode (%d)",
85982a32 184 enc->name[0], (U8) s[slen], code);
185 }
186 goto ENCODE_SET_SRC;
af1f55d9 187 }else if (check &
188 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
0ab8f81e 189 SV* perlqq =
fc17bd48 190 sv_2mortal(newSVpvf("\\x%02" UVXf, s[slen]));
85982a32 191 sdone += slen + 1;
192 ddone += dlen + SvCUR(perlqq);
193 sv_catsv(dst, perlqq);
194 } else {
195 sdone += slen + 1;
0ab8f81e 196 ddone += dlen + strlen(FBCHAR_UTF8);
197 sv_catpv(dst, FBCHAR_UTF8);
85982a32 198 }
aa0053b7 199 }
b2704119 200 }
85982a32 201 /* settle variables when fallback */
b0b300a3 202 d = (U8 *)SvEND(dst);
0ab8f81e 203 dlen = SvLEN(dst) - ddone - 1;
204 s = (U8*)SvPVX(src) + sdone;
b2704119 205 slen = tlen - sdone;
206 break;
2f2b4ff2 207
85982a32 208 default:
209 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
210 code, (dir == enc->f_utf8) ? "to" : "from",
211 enc->name[0]);
212 return &PL_sv_undef;
aa0053b7 213 }
85982a32 214 }
215 ENCODE_SET_SRC:
6d1c0808 216 if (check && !(check & ENCODE_LEAVE_SRC)){
217 sdone = SvCUR(src) - (slen+sdone);
85982a32 218 if (sdone) {
219 sv_setpvn(src, (char*)s+slen, sdone);
aa0053b7 220 }
85982a32 221 SvCUR_set(src, sdone);
2f2b4ff2 222 }
85982a32 223 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
0ab8f81e 224
85982a32 225 SvCUR_set(dst, dlen+ddone);
226 SvPOK_only(dst);
0ab8f81e 227
39cf9a5e 228#if ENCODE_XS_PROFILE
229 if (SvCUR(dst) > SvCUR(src)){
85982a32 230 Perl_warn(aTHX_
231 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
232 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
233 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
39cf9a5e 234 }
3c49ab08 235#endif
0ab8f81e 236
85982a32 237 ENCODE_END:
0b3236bb 238 *SvEND(dst) = '\0';
aa0053b7 239 return dst;
2f2b4ff2 240}
241
50d26985 242MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
2f2b4ff2 243
244PROTOTYPES: ENABLE
245
246void
0a95303c 247Method_name(obj)
248SV * obj
249CODE:
85982a32 250{
251 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
252 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
253 XSRETURN(1);
254}
0a95303c 255
256void
b2704119 257Method_decode(obj,src,check = 0)
2f2b4ff2 258SV * obj
259SV * src
b2704119 260int check
2f2b4ff2 261CODE:
aae85ceb 262{
85982a32 263 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
264 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
265 SvUTF8_on(ST(0));
aae85ceb 266 XSRETURN(1);
267}
268
269void
85982a32 270Method_encode(obj,src,check = 0)
aae85ceb 271SV * obj
85982a32 272SV * src
273int check
aae85ceb 274CODE:
275{
85982a32 276 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
277 sv_utf8_upgrade(src);
278 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
aae85ceb 279 XSRETURN(1);
280}
281
0ab8f81e 282void
283Method_needs_lines(obj)
284SV * obj
285CODE:
286{
287 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
288 ST(0) = &PL_sv_no;
289 XSRETURN(1);
290}
291
292void
293Method_perlio_ok(obj)
294SV * obj
295CODE:
296{
297 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
10c5ecbb 298 /* require_pv(PERLIO_FILENAME); */
299
300 eval_pv("require PerlIO::encoding", 0);
301
302 if (SvTRUE(get_sv("@", 0))) {
0ab8f81e 303 ST(0) = &PL_sv_no;
10c5ecbb 304 }else{
305 ST(0) = &PL_sv_yes;
0ab8f81e 306 }
307 XSRETURN(1);
308}
309
67e989fb 310MODULE = Encode PACKAGE = Encode
2c674647 311
312PROTOTYPES: ENABLE
313
67e989fb 314I32
2c674647 315_bytes_to_utf8(sv, ...)
85982a32 316SV * sv
317CODE:
318{
319 SV * encoding = items == 2 ? ST(1) : Nullsv;
0ab8f81e 320
85982a32 321 if (encoding)
322 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
323 else {
324 STRLEN len;
325 U8* s = (U8*)SvPV(sv, len);
326 U8* converted;
327
328 converted = bytes_to_utf8(s, &len); /* This allocs */
329 sv_setpvn(sv, (char *)converted, len);
330 SvUTF8_on(sv); /* XXX Should we? */
331 Safefree(converted); /* ... so free it */
332 RETVAL = len;
333 }
334}
335OUTPUT:
336 RETVAL
2c674647 337
67e989fb 338I32
2c674647 339_utf8_to_bytes(sv, ...)
85982a32 340SV * sv
341CODE:
342{
343 SV * to = items > 1 ? ST(1) : Nullsv;
344 SV * check = items > 2 ? ST(2) : Nullsv;
345
346 if (to) {
347 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
348 } else {
349 STRLEN len;
350 U8 *s = (U8*)SvPV(sv, len);
351
352 RETVAL = 0;
353 if (SvTRUE(check)) {
354 /* Must do things the slow way */
355 U8 *dest;
356 /* We need a copy to pass to check() */
0ab8f81e 357 U8 *src = (U8*)savepv((char *)s);
85982a32 358 U8 *send = s + len;
359
360 New(83, dest, len, U8); /* I think */
361
362 while (s < send) {
363 if (*s < 0x80){
364 *dest++ = *s++;
365 } else {
366 STRLEN ulen;
367 UV uv = *s++;
368
369 /* Have to do it all ourselves because of error routine,
370 aargh. */
371 if (!(uv & 0x40)){ goto failure; }
372 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
373 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
374 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
375 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
376 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
377 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
378 else { ulen = 13; uv = 0; }
87714904 379
85982a32 380 /* Note change to utf8.c variable naming, for variety */
381 while (ulen--) {
0ab8f81e 382 if ((*s & 0xc0) != 0x80){
383 goto failure;
85982a32 384 } else {
385 uv = (uv << 6) | (*s++ & 0x3f);
386 }
87714904 387 }
67e989fb 388 if (uv > 256) {
389 failure:
85982a32 390 call_failure(check, s, dest, src);
391 /* Now what happens? */
67e989fb 392 }
393 *dest++ = (U8)uv;
85982a32 394 }
395 }
396 } else {
397 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
2c674647 398 }
85982a32 399 }
400}
401OUTPUT:
402 RETVAL
2c674647 403
2c674647 404bool
b2704119 405is_utf8(sv, check = 0)
4411f3b6 406SV * sv
b2704119 407int check
85982a32 408CODE:
409{
410 if (SvGMAGICAL(sv)) /* it could be $1, for example */
411 sv = newSVsv(sv); /* GMAGIG will be done */
412 if (SvPOK(sv)) {
413 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
414 if (RETVAL &&
415 check &&
416 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2c674647 417 RETVAL = FALSE;
85982a32 418 } else {
419 RETVAL = FALSE;
420 }
421 if (sv != ST(0))
422 SvREFCNT_dec(sv); /* it was a temp copy */
423}
424OUTPUT:
425 RETVAL
2c674647 426
427SV *
4411f3b6 428_utf8_on(sv)
85982a32 429SV * sv
430CODE:
431{
432 if (SvPOK(sv)) {
433 SV *rsv = newSViv(SvUTF8(sv));
434 RETVAL = rsv;
435 SvUTF8_on(sv);
436 } else {
437 RETVAL = &PL_sv_undef;
438 }
439}
440OUTPUT:
441 RETVAL
2c674647 442
443SV *
4411f3b6 444_utf8_off(sv)
85982a32 445SV * sv
446CODE:
447{
448 if (SvPOK(sv)) {
449 SV *rsv = newSViv(SvUTF8(sv));
450 RETVAL = rsv;
451 SvUTF8_off(sv);
452 } else {
453 RETVAL = &PL_sv_undef;
454 }
455}
456OUTPUT:
457 RETVAL
458
85982a32 459int
460DIE_ON_ERR()
461CODE:
462 RETVAL = ENCODE_DIE_ON_ERR;
463OUTPUT:
464 RETVAL
465
0ab8f81e 466int
85982a32 467WARN_ON_ERR()
468CODE:
469 RETVAL = ENCODE_WARN_ON_ERR;
470OUTPUT:
471 RETVAL
472
473int
474LEAVE_SRC()
475CODE:
476 RETVAL = ENCODE_LEAVE_SRC;
477OUTPUT:
478 RETVAL
479
480int
481RETURN_ON_ERR()
482CODE:
483 RETVAL = ENCODE_RETURN_ON_ERR;
484OUTPUT:
485 RETVAL
486
487int
488PERLQQ()
489CODE:
490 RETVAL = ENCODE_PERLQQ;
491OUTPUT:
492 RETVAL
493
494int
af1f55d9 495HTMLCREF()
496CODE:
497 RETVAL = ENCODE_HTMLCREF;
498OUTPUT:
499 RETVAL
500
501int
502XMLCREF()
503CODE:
504 RETVAL = ENCODE_XMLCREF;
505OUTPUT:
506 RETVAL
507
508int
85982a32 509FB_DEFAULT()
510CODE:
511 RETVAL = ENCODE_FB_DEFAULT;
512OUTPUT:
513 RETVAL
514
515int
516FB_CROAK()
517CODE:
518 RETVAL = ENCODE_FB_CROAK;
519OUTPUT:
520 RETVAL
521
522int
523FB_QUIET()
524CODE:
525 RETVAL = ENCODE_FB_QUIET;
526OUTPUT:
527 RETVAL
528
529int
530FB_WARN()
531CODE:
532 RETVAL = ENCODE_FB_WARN;
533OUTPUT:
534 RETVAL
535
536int
537FB_PERLQQ()
538CODE:
539 RETVAL = ENCODE_FB_PERLQQ;
540OUTPUT:
541 RETVAL
2c674647 542
af1f55d9 543int
544FB_HTMLCREF()
545CODE:
546 RETVAL = ENCODE_FB_HTMLCREF;
547OUTPUT:
548 RETVAL
549
550int
551FB_XMLCREF()
552CODE:
553 RETVAL = ENCODE_FB_XMLCREF;
554OUTPUT:
555 RETVAL
556
33af2bc7 557BOOT:
558{
85982a32 559#include "def_t.h"
e7cbefb8 560#include "def_t.exh"
33af2bc7 561}